TAILIEUCHUNG - Parallel Programming: for Multicore and Cluster Systems- P14

Parallel Programming: for Multicore and Cluster Systems- P14: Innovations in hardware architecture, like hyper-threading or multicore processors, mean that parallel computing resources are available for inexpensive desktop computers. In only a few years, many standard software products will be based on concepts of parallel programming implemented on such hardware, and the range of applications will be much broader than that of scientific computing, up to now the main application area for parallel computing | Information Exchange 121 Pi xi X2 Xp Pi xT P2 Q P2 XT scatter . Pp Q Pp xp To perform the scatter each processor explicitly calls a scatter operation and specifies the root processor as well as a receive buffer. The root processor additionally specifies a send buffer in which the data blocks to be sent are provided in rank order of the rank i 1 . p. Multi-broadcast The effect of a multi-broadcast operation is the same as the execution of several single-broadcast operations one for each processor . each processor sends the same data block to every other processor. From the receiver s point of view each processor receives a data block from every other processor. Different receivers get the same data block from the same sender. The operation can be illustrated as follows P1 X1 Pi X1 II X2 II II Xp P2 X2 P2 X1 II X2 II II Xp multi-broadcast . . Pp Xp Pp X1 II X2 II II Xp In contrast to the global operations considered so far there is no root processor. To perform the multi-broadcast each processor explicitly calls a multi-broadcast operation and specifies a send buffer which contains the data block as well as a receive buffer. After the completion of the operation the receive buffer of every processor contains the data blocks provided by all processors in rank order including its own data block. Multi-broadcast operations are useful to collect blocks of an array that have been computed in a distributed way and to make the entire array available to all processors. Multi-accumulation The effect of a multi-accumulation operation is that each processor executes a single-accumulation operation . each processor provides for every other processor a potentially different data block. The data blocks for the same receiver are combined with a given reduction operation such that one reduced data block arrives at the receiver. There is no root processor since each processor acts as a receiver for one accumulation operation. The effect of the operation with addition as .

• Phần cứng
• sửa chữa máy tính
• sổ tay phần cứng
• MainBoard
• thiết bị máy tính
• thiết bị mạng
• lắp ráp máy tính
• Phần cứng máy tính
• lỗi máy tính
• hệ thống mạch máy tính
• điện máy tính
• Tài liệu máy tính